ultraqa

Automated test fixing uses a goal-driven QA cycling loop that repeatedly runs verification, diagnoses root causes, applies corrections, and retries until checks pass. You select a goal like tests, build, lint, or typecheck to start the iterative diagnose-and-fix loop.

A self-healing QA cycle is an iterative loop that detects failures, routes analysis to an architect step, applies fixes via an executor step, and repeats. It enforces a maximum of five cycles and tracks progress in a state file for resumable behavior.

Yes, you can run linting and typecheck fixes separately by selecting a specific target. The goal-driven QA cycles support distinct selectable goal types including tests, build, lint, and typecheck, allowing you to target only the failing checks you need.

Yes, the automated debugging loop stops if the same failures keep repeating. It enforces early exit conditions on repeating failures and limits the entire process to a maximum of five cycles to prevent infinite loops.

You resume an interrupted CI fix cycle through persisted state tracking. The process saves progress automatically in a local state file, allowing the deterministic success or failure conditions to resume from where the previous iteration left off.

An automated test runner exits before reaching a passing state when it hits the maximum of five cycles or encounters repeating failures. These deterministic exit conditions prevent infinite loops when analysis and remediation cannot resolve the root cause.